What 10,765 Hires Revealed About Resume Keywords and Performance

Across 10,765 hires at a Fortune 500 insurance carrier, researchers tested whether resume keywords predict job performance. They parsed 8,181 unique skills from applicant tracking system profiles and tested the 3,597 that had enough data to measure. After Bonferroni correction, the strictest standard for multiple comparisons, zero keywords predicted production. Thirty were significantly anti-predictive. The median keyword was associated with about 25% lower odds of producing, and more keywords on a resume correlated with lower production, not higher.

Source: "Decision Traces," Saad Bin Shafiq, NODES, 2026. N=10,765 agents hired 2022 to 2025 at a Fortune 500 insurance carrier, analyzed inside the carrier's VPC. Read it on arXiv.

The test: every keyword, against real outcomes

The dataset held 8,181 unique parsed skills, a mean of 27.2 per agent, across 10,362 evaluable hires. Every skill that appeared on at least five profiles was tested for its association with production, which yielded 3,597 testable keywords. The team then applied Bonferroni correction across all 3,597 simultaneous tests.

The result was clean and one-directional. Zero keywords significantly predicted production. Thirty were significantly anti-predictive. Across all tested keywords, 70.2% had an odds ratio below 1, meaning listing them was associated with a lower chance of producing. The median odds ratio was 0.749.

The most anti-predictive keywords

The strongest positive keyword was Weddings (odds ratio 1.857), and even that did not survive correction. Positive keywords that looked significant were driven by three to five people and were not statistically reliable.

Standard insurance filters did worse than no filter

The team then checked the six categories the industry actually screens on:

Both held up within individual hiring years, so they are not a cohort artifact. The two signals the industry treats as table stakes were associated with lower production.

The candidates a keyword filter throws away

The carrier had 677 agents with no traditional keywords at all who scored 75 or above on behavioral assessment. They produced at 33.7%, above the overall population rate. Agents who had every keyword present produced at only 26.3%. The keyword filter was selecting in the wrong direction. See what that cost in dollars.

What this does not say

This is one carrier and one role type, insurance sales. The decision-trace method generalizes. The specific keyword rankings may be carrier-specific. The finding is a reason to test your own keywords against your own outcomes, not to assume the same numbers will appear in your data.

Frequently asked questions

Do resume keywords predict job performance? In this study of 10,765 insurance hires, no. Of 3,597 testable keywords, none predicted production after correcting for multiple comparisons, and 30 were anti-predictive.

What is Bonferroni correction and why does it matter here? It is the strictest adjustment for testing many hypotheses at once. Applied to 3,597 keyword tests, it removes any keyword that looks predictive purely by chance. Zero keywords survived it.

Does having more skills on a resume help? No. Agents with 0 to 10 parsed skills produced at 36.1%, while agents with 51 or more produced at 22.7%.

Is this finding specific to insurance? The data comes from insurance agent hiring. The method applies to any role where screening inputs can be connected to production outcomes.

Related reading

• Why ATS keywords fail to predict performance
• The $17.7M cost of a single hiring filter
• Decision traces, explained

Want to know which of your screening rules are anti-predictive? Book a 30-minute walkthrough.